Load a Snapshot and Add a Pillar: LoadOptimizationFromJsonAndReassignPillar

This example loads a saved optimization state from myopti.json.bz2 and adds one pillar node (a fixed appointment) to it before the next run. It shows how a planned state can take a new appointment with a strict time window without rebuilding the model.

Links

• Example source (GitHub):
  LoadOptimizationFromJsonAndReassignPillar.java
• Overview of all load and save examples in this package:
  LoadAndSaveOptimization.md
• Pillar nodes in general:
  Pillar.md

Before you run it

The example reads myopti.json.bz2 from the working directory. Run SaveOptimizationToJsonExample first, in the same working directory; it writes that file at the end of its run.

If the file is missing, the example prints:

JsonFile not existing. Run example 'SaveOptimizationToJsonExample.java' first.

followed by the stack trace of the FileNotFoundException, and does not start a run.

Execution flow

example() does the following:

1. Sets the license via ExampleLicenseHelper.setLicense(this);
2. Loads the snapshot into this optimization:
   • IOptimizationImporter importer = new OptimizationJSONImporter();
   • importer.update(jsonFile, opti);
3. Adds the pillar through reassignment:
   • this.addReassignNodes(getAdditionalNodes());
4. Sets the properties for the new run via setProperties().
5. Starts the run with startRunAsync().get() and prints the result in onAsynchronousOptimizationResult(...).

The added pillar

getAdditionalNodes() creates one PillarTimeWindowGeoNode:

• id: Koeln-Pillar
• position: 50.9333, 6.95 (the coordinates the other examples use for Koeln)
• opening hours: March 7, 2020, 08:00 to 17:00 (Europe/Berlin)
• visit duration: 20 minutes

IPillarNode koeln =
    new PillarTimeWindowGeoNode("Koeln-Pillar", 50.9333, 6.95, weeklyOpeningHours.get(0), visitDuration);

The code does not attach the pillar to a resource (there is no attachResource(...) call, unlike in PillarExample).

A pillar is a fixed appointment: its time window must be met. addReassignNodes(...) hands the pillar to the loaded optimization, so the next run plans it together with the elements from the snapshot.

Solver properties

setProperties() registers:

• JOptExitCondition.JOptGenerationCount = 2000
• JOpt.Algorithm.PreOptimization.SA.NumIterations = 100000
• JOpt.Algorithm.PreOptimization.SA.NumRepetitions = 1
• JOpt.NumCPUCores = 4

Related examples

• LoadOptimizationFromJsonAndReassignNodes: the same pattern with three ordinary time-window nodes instead of a pillar.
• LoadOptimizationToFreshRunExample: loads the snapshot but ignores its solution.
• PillarExample: pillars attached to resources, built from scratch.

How to run

1. Run the main(String[] args) method of SaveOptimizationToJsonExample.
2. Run the main(String[] args) method of LoadOptimizationFromJsonAndReassignPillar in the same working directory.